# Changeset 532

Ignore:
Timestamp:
Mar 2, 2006, 6:44:48 PM (17 years ago)
Message:

modified implementation of Parson weighted - now in line with other weighted statistics.

Location:
trunk/lib/statistics
Files:
4 edited

Unmodified
Removed
• ## trunk/lib/statistics/SNR.h

 r529 /// /// Class for score based on signal-to-noise ratio (SNR).  Also /// sometimes referred to as Golub score. The score is the ratio /// between difference in mean and the sum of standard deviations /// for two groups.  \f$\frac{ \frac{1}{n_x}\sum x_i - /// \frac{1}{n_y}\sum y_i }{ \sigma_x + \sigma_y} \f$ where \f$/// \sigma \f$ is standard deviation. /// /// @brief Class for score based on signal-to-noise ratio (SNR). /// /// Also /// sometimes referred to as Golub score. The score is the ratio /// between difference in mean and the sum of standard deviations /// for two groups.  \f$\frac{ \frac{1}{n_x}\sum x_i - /// \frac{1}{n_y}\sum y_i }{ \sigma_x + \sigma_y} \f$ where \f$/// \sigma \f$ is standard deviation. /// class SNR : public Score {
• ## trunk/lib/statistics/Score.h

 r529 /// be used when two-tailed test is wanted. /// /// @return statistica. /// /// @param target vector of targets (most often +1 -1) /// @param value vector of the values /// virtual double score(const classifier::Target& target, /// is wanted. /// /// @return statistica (weighted version) /// /// @param target is +1 or -1 /// @param value vector of the values /// @param weight vector of accompanied weight to the values /// virtual double score(const classifier::Target& target, /// calculated. Absolute mode should be used when two-tailed test /// is wanted. /// /// @return statistica (weighted version) /// /// @param target is +1 or -1 /// @param value vector of the values /// @param weight vector of accompanied weight to the values /// inline double
• ## trunk/lib/statistics/tScore.cc

 r529 statistics::Averager positive; statistics::Averager negative; dof_=target.size()-2; for(size_t i=0; i
• ## trunk/lib/statistics/tScore.h

 r529 /// Weighted version of t-Score @return t-score if absolute=true /// absolute value of t-score is returned. /// @todo Peter: This is probably /// not implemented as it should be! /// /// @todo document /// double score(const classifier::Target& target, private: double t_; int dof_; double dof_; };
Note: See TracChangeset for help on using the changeset viewer.